Fig. 3

Epigenomic annotation and principal component analysis of the K562 regulatory modules. a A heatmap showing the fraction of regions using the same modules (rows) overlapping the specific annotations (columns). The modules (rows) are ordered as in Fig. 2. The top dendrogram was computed by applying hierarchical clustering on the fraction matrix with Pearson correlation distance and average linkage. b Principal component analysis (PCA) was performed on the factor-module matrix. The PCA loadings of the modules for the first four PCs. c The scatter plot shows the TF binding site counts projected on the principal components (PCs) 1 and 2. The percentage labels in the x and y axes represent the percentages of variance explained by the PCs 1 and 2, respectively. d Similar to c. The TF binding site counts projected on the PC3 and PC4. CS: chromatin state; T: transcribed region; R: repressed region; PF: promoter flanking region; TSS: promoter region including TSS; E: enhancer region; WE: weak enhancer or open chromatin region